Years ago, no high-end PC was complete without a couple of boot drives configured in RAID 0. Today, things are a bit different. SSDs are available for reasonable prices, in high capacities, and with even higher speeds. But you don't have to throw RAID away. There are still a ton of benefits to using SSDs in a RAID array between redundancy, parity, and with the right RAID controller, even higher speeds. Still, SSDs and HDDs aren't built equally, and there are some common mistakes you should navigate around if you're configuring SSDs into a RAID array.

5 Using mismatched SSDs

Be careful of bottlenecking your speed or capacity

There isn't a strict rule against mixing and matching drives in a RAID array, and in some cases, you may even want to use different drives from different brands. Unless you're volume-buying batches of SSDs and looking to weed out manufacturing errors, though, it's best to use identical SSDs in your RAID array. The performance of your array will always be limited by the slowest drive, which could reduce your capacity, reduce your speeds, or both. In a RAID 1 configuration, for example, your slowest drive will be the limiting factor for speeds because data is mirrored across the drives. In a RAID 0 configuration, capacity is the limiting factor, as you're only able to stripe data up to the capacity of the lowest drive in your array.

It's not inherently bad to mix and match drives, but it's important to understand how that can impact your performance and capacity when creating an array. Assuming you're setting up an array on your PC or in a home NAS, you'll almost always want the same drive in the same capacity. It doesn't need to be from the same brand, but you'll want to make sure the capabilities of each SSD in your array are evenly matched. The only time this is a bad idea is if there happens to be a bad batch of SSDs. If you're unfortunate enough to order a bunch of drives with some sort of manufacturing defect, you may have to deal with several issues on your drives back-to-back. That shouldn't happen often, though.

4 Using the wrong drives

Enterprise or consumer?

You can create a RAID array with no more than two drives and a dream, but it's important to consider what you're setting up the RAID array for if you're after the best performance and longevity. If you're setting up a small array in your personal rig, you should be fine with a regular consumer SSD. Particularly in RAID 0 or 1 configurations with only a few drives, consumer drives are cheaper than enterprise options and still provide great speeds. They're built for bursts of speed for a single user, rather than multiple users, and steady performance 24 hours per day.

Some RAID arrays are built for a different purpose, be it long-term backup or a home server, and for these types of applications, you'll want to use enterprise SSDs. Enterprise SSDs are significantly more expensive than their consumer counterparts, but they're also designed to run high-intensity workloads around the clock. A typical consumer SSD comes with a TBW rating, or Terabytes Written, to show its endurance. Enterprise SSDs, on the other hand, come with DWPD number, or Drive Writes Per Day, which looks at how many times you can fully fill up the drive each day over its warrantied lifespan.

If you're just putting together a small array, anything other than consumer SSDs is probably a waste of money. And if you're building some more intensive, going with anything other than enterprise drives is going to create headaches down the line.

3 Running at capacity

The 80/20 rule applies to SSDs in RAID, too

Running even a single SSD at full capacity is a bad idea, and it's even worse when multiple SSDs are configured in a RAID array. As you near the full capacity of your SSD, the performance starts to degrade. In a RAID array, that performance degradation is almost always exaggerated. As is always the case in a RAID array, the weakest drive will become a bottleneck. And since you'll be running at least two SSDs at full capacity in most RAID configurations, you'll be limited by the performance hit the weakest of the drives has.

You need to keep some free space in your RAID array for the best performance with SSDs. That's due to the garbage collection process of an SSD, where empty pages are returned to the drive as you write new data. Although important for a single SSD, garbage collection is even more important with a RAID array, as some RAID controllers and levels don't support TRIM, and SSDs have larger issues with write amplification than HDDs do. There's no hard rule for how much free space you need, but it's usually somewhere around 15% to 20% where you start to see performance degradation.

2 Configuring a write-heavy array without the proper drives

Parity means more writes to your SSDs

If you're configuring a RAID array with parity, and you're using SSDs, you need to use enterprise drives. Unlike HDDs, where the only thing limiting their lifespan are mechanical parts, SSDs have a set limit of writes before the cells are ready for the trash. If you're using a RAID level with parity, like RAID 4, 5, or 6, you're introducing more writes to your array. In the context of RAID, parity means that the array can be rebuilt if a drive fails. It's not the same as mirroring. RAID 1 uses mirroring, but it doesn't use parity. RAID levels with parity either have a dedicated parity drive or the data is distributed across the array.

Parity increases the amount of data written to your RAID array. Parity bits that are used to rebuild your array in the event of a drive failure take up extra space, and they increase the number of writes you have to make to the drive. This isn't something you have to worry about with RAID 0 or 1, thankfully.

That doesn't mean you should avoid RAID 4, 5, or 6 entirely if you're using SSDs. That's not the case at all. It's just important to keep in mind that these configurations will degrade your drives faster than RAID 0 or 1. If you're planning on putting together a RAID 4, 5, or 6 array with SSDs, make sure to use enterprise drives. At least then you'll have a better idea of when a drive should fail, and you can plan accordingly to rebuild your data.

1 Ignoring your firmware

Controller and storage

There are two types of people in this world: those who install new updates right away, and those who adopt an "if it ain't broke, don't fix it mentality." Both are valid, and there are reasons you might want to stay on older firmware for your RAID controller. Regardless, it's important to at least pay some mind to the firmware on both your SSDs and your RAID controller. For your RAID controller, things are straightforward. Bugs in the firmware can lead to lost data down the line, so you'll want to figure out a firmware update process so you can easily flash new firmware when needed.

You don't need the latest firmware for your SSDs, but it's still a good idea to get everything up to the same spot before configuring your array. SSDs are much more complex than HDDs, packing in I/O, controllers, extra cache, and even a small processor, and that's before getting to the actual flash chips. SSDs can have bugs that lead to lost data, and they don't always surface right away. Some drives will fail due to a bug after a certain number of hours, or a certain numbers of bytes written. Firmware updates address these bugs.

SSDs make for a killer array

Although RAID isn't nearly as popular as it used to be for performance in the time of high-speed, high-capacity SSDs for a reasonable price, there are still a ton of advantages to using SSDs in your RAID array. You won't always see a performance benefit, but RAID still gives you parity and redundancy options for storing a ton of data.